Segmentation of dermatological ulcers using clustering of color components

Chronic wounds, lesions, or ulcers due to venous insufficiency and other conditions typically have a nonuniform mixture of red granulation, yellow fibrin (slough), black necrotic eschar (scar), and white hyperkeratotic tissue (callous). Such a red-yellow-black-white (RYKW) model is used by clinicians as a descriptive tool. To facilitate the analysis of the tissue composition of a lesion, we propose color imaging and image processing methods. Methods based on clustering of color components in hue-saturation histograms and mathematical morphology are proposed for the segmentation of a given image into regions corresponding to red, yellow, black, and white tissue. Tests with 172 images indicated an average Jaccard coefficient of 0.56 with a standard deviation of 0.22 between the lesion area obtained computationally and the same lesion region manually delineated by a dermatologist. More importantly, a low average root-mean-squared error of 4% with a standard deviation of 5% was obtained between the tissue composition of lesions estimated using the proposed segmentation method and manual analysis.